Internal PSU Repair

[Steve Walker]

Hi gang -

The internal PSU in my trusty Netgear FS308 ethernet switch has died during a period of mains instability (but if it couldn't cope then it was probably nine-tenths gone anyway). Being a silly, sentimental old sod I'd like to repair it but I've run out of skill pretty fast...

It's a Delta Electronics ADP-30JP, outputting 5v @ 30w/6a - I'm guessing that will be a 'regulated' output. Dimensions are 100mm L, 45mm W, 30mm H. (Imageshack pics @

, , , have closely examined the unit and can see no obvious scorching, leakage or other damage. There's a cylindrical black 2a mains fuse on the board (bottom left in image 2) but that has good continuity. When powered-up it appears to output approx 0.3v dc, and the capacitors are very healthy (discovered this when holding the disconnected unit on the palm of my hand).

If anyone has any ideas for further testing or replacement, or has a spare one.. ?

Thanks in advance -

[Dave Plowman (News)]

Just because the caps hold a charge doesn't mean they're ok. You need something like an ESR meter to test them properly - or by substitution. But note you need low ESR types. You could check the voltage at the output of the mains rectifier as a starting point, though. Other things you could check easily are all the various diodes.

But it might be easier to just replace the PS - maybe with an external one.

[Andrew Gabriel]

A new 8 port 10/100 ethernet switch will cost less than £10. You are very unlikely to find a suitable replacement PSU for that amount.

If you are particularly wedded to that model (which I suspect is long obsolete), then buy one off eBay (but eBay is always well over-priced for ethernet switches for some reason - the used FS308 on there at the moment "buy it now" works out at more than I paid for a brand new 8 port gigabit switch).

[dennis]

switch).

[Mike Harrison]

A very common failure mode in SMPS's is an electrolytic cap in the startup circuit, causing the PSU to fail to restart after a power interruption. This is often mis-attributed to mains problems/spikes etc. as the fault only exhibits after a switch-off, when it then fails to re-start - the problem may actually have been there unnoticed for years if it has never been switched off.

Look for a small, low-voltage (

[NT]

rote:

p circuit, causing the PSU

ed to mains problems/spikes

re-start - the problem may

deg.C temp rating).

ly hurt for many minutes

either diagnose the fault or youre wasting your time.

5v 6A =3D 30w. Over 1 yr that's 263kWh costing =A326.30 @ 10p/unit or =A331.53 @ 12p/unit. Over 5 years that's =A3131 or =A3157.

NT

[Steve Walker]

Thanks Mike, that was an amazing piece of psychic diagnosis - salvaged a 50v capacity from elsewhere, and it's running fine again!

[Steve Walker]

I should clarify that it doesn't actually use 30w. Factory data for flat-out all-ports usage is 15.2w, the rest is presumably safety margin. That sort of build quality, along with the steel case and DIP-switched manually-configurable duplexing, is why I'm so fond of the old thing.

Thanks to all for comments, but especially to Mike Harrison for the fix tip.

[Clint Sharp]

The little one in the corner of the L shaped heatsink yes? Really common failure, usually 22uF - 47uF, it smooths the voltage used to run the switcher IC (UC3844 probably) until the PSU is up and self sustaining, once it's running it will run almost forever until you power it down and then it won't restart. It's a very useful bit of knowledge because it's such a common fault.

[Steve Walker]

You're just showing off now - yes, it was that one.... :o)

[Clint Sharp]

Hehe, sorry. Spent lots of time repairing power supplies and other PC boards when it was worth the effort to do. You get to learn the 'standard' faults pretty quickly and the first rule of repairing PSUs is that it's almost always the capacitor with a 4 and a 7 in the value if the fuse is OK.

[Graham.]

And if the fuse has gone, (literally as in vaporised) then it'll be your input bridge reccy s/c.

Of course at the other end of the scale you get faults that take out multiple components in the control circuitry including the opto, control IC, chopper, and several diodes etc and unless you replace all of them the thing just avalanches and writes off an entire CPC order in an instant.

[Andrew Gabriel]

Switching transistor can do that too, and sometimes ends up with a cracked case and a trail of flash burn ejected from it. Also had one where the main smoothing capacitor exploded. The SMPSU actually carried on working, with a stonking 100Hz ripple on the output, which that particular load (a light) didn't care about, except it ran a bit dimmer.

Another rather bizzare fault I've seen a couple of times in old SMPSUs is the glue used to bond the large capacitor(s) to the PCB eventally turns into a conductor, and flashes across, which I've repaired by carefully scraping it off. Initially thought capacitors where leaking or failed, but actually they were fine in both cases.

[Jules]

Had one in a piece of Sun equipment once a bit like that - the cascade of failures as one thing took out another, all of which happened in the blink of an eye, was quite beautiful in a way.

Cue many, many failed components, some outright blown apart, with the final stage before the main fuse* did its job being a handful of PCB traces getting hot enough to lift right off the board.

• the equipment being there to protect the fuse at all costs, of course.

It was damaged enough that it was hard to speculate what the primary failure actually was - although it was quite reassuring that the backup supply did its thing, and everything kept running even after the loud crackling noise and the room slowly filled with the smell of burning :-)

cheers

Jules

[gorgoth]

you just saved my old switch as well thank you

[marman_66]

Thank you very much!! Saved 3 network switch from trash-bin!

[amin]

replying to Steve Walker, amin wrote: you must change c6

[The Natural Philosopher]

No. It almost certainly is R13.

[F]

It was R5 here.

[Sam Plusnet]

Quarter to six Shirley?